(no subject)

From: David Rogers (wantye_at_gmail.com)
Date: Mon Aug 22 2011 - 20:25:04 CDT

On Thu, Aug 18, 2011 at 5:59 AM, Andrea Holfelder
<andreaholfelder_at_gmx.com> wrote:
> Is it possible to create an RHDO lattice as a PBC unit cell for MD
> simulations in NAMD ?
> I have read/heard that RHDO lattice is better and faster than cubic etc.
> units.
> ----
> Andrea Holfelder

  NAMD does triclinic boxes. According to H. Bekker's Ph.D. thesis
(Molecular Dynamics Simulation Methods Revised), any space-filling
shape is identical to a triclinic box:

'In molecular simulations with periodic boundary conditions the
computational box
used, may have five different shapes: triclinic, the hexagonal prism,
two types of
dodecahedrons, and the truncated octahedron. In this chapter we show that every
molecular simulation, irrespective of the shape of the initial
computational box, can
be done as a simulation in one of the other ones, i.e. we show that in
a preprocessing
phase a simulation formulated in one particular box can be transformed into a
simulation in another box such that the simulation in the new box is
exactly identical
to the simulation in the original one. This means that every molecular
may be done in the same type of box. Because the triclinic box is the
easiest one to
implement, we pay special attention on how to transform the other four
box types into
triclinic boxes. As a consequence, simulations in the often used
truncated octahedron
are superfluous; they may be done in a much simpler way in a triclinic box.'

  If you want another shape (with the same lattice vectors), you can
take the DCD (make sure you saved the box size/shape in the dcd, this
is a namd config option somewhere), convert it to a gromacs trr using
VMD, then run trjconv to wrap the molecules appropriately.

~ David M. Rogers
  Postdoc, Susan Rempe Lab,
  Sandia National Labs

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:57:38 CST